Double-tracer-agent PET separation method based on multi-task learning three-dimensional convolutional encoding and decoding network

Abstract

The invention discloses a double-tracer-agent PET separation method based on a multi-task learning three-dimensional convolutional encoding and decoding network. According to the method, a signal separation task of double-tracer-agent PET is converted into two single tracer agent PET reconstruction problems based on a multi-task learning three-dimensional convolutional encoding and decoding network; training data and label values are inputted into a built neural network, the relation between a mixed double-tracer-agent PET sinusoidal image sequence and PET concentration image sequences of twosingle tracers is learned, and therefore signal separation of double-tracer-agent PET is completed.

Description

technical field [0001] The invention belongs to the technical field of PET imaging, and in particular relates to a dual-tracer PET separation method based on a multi-task learning three-dimensional convolution encoding and decoding network. Background technique [0002] Positron emission tomography (PET) is a diagnostic functional imaging technique that can detect physiological and chemical activities in the human body. Using this technology, it can obtain the glucose metabolism, blood flow and hypoxia and other diseases in target tissues in the human body. related physiological indicators. The principle of PET is: a compound (tracer) labeled with a positron isotope enters the body through injection or oral administration, and according to the needs of the internal physiological or pathological activities of the human body, the tracer gathers in the part where the demand for this compound is high; when the radioactive nucleus When the element decays, it will emit positrons....

AI Technical Summary

Problems solved by technology

In order to save the scanning time and cost of patients, how to deal with single-scan dual-tracer PET signals has become a key technology that needs to be solved urgently; since the annihilation of different radionuclides produces a pair of 511ekV gamma photons, so for the signal source Since it is difficult to distinguish which tracer, it is impossible to know the spatial and temporal distribution of the two tracers and the monitored vital activity status

[0004] At present, the dual-tracer PET separation methods can be mainly divided into: (1) Using tracer prior information and interval injection combined with mathematical models to distinguish the signals of different tracers, the commonly used mathematical models in this type of method include compartment model method, base tracing method, and signal epitaxy method, etc.; taking the signal epitaxy method as an example, it injects two tracers successively to obtain a time-activity curve of the mixed tracer with a time interval, and then uses a mathematical model to fit and show The non-overlapping part of the tracer-time-activity curve, and then extrapolating the overlapping part of tracer I and tracer II, can complete the separation of the time-activity curves of the two tracers; this type of method has the following problems: 1 .The single tracer in the mixed tracer is required to have different half-lives or different radioactive isotopes, which reduces the practical feasibility of the method; 2. A pre-constructed prior mathematical model is required, and the prior model has no effect on the new Tracers may not be applicable; 3. Requires interval between injections, prolonging scan time

(2) Adopt the double tracer separation technology of prompt Gammas. This method requires one of the tracers to emit one more high-energy gamma ray. According to this characteristic, the signal separation of the double tracers can be realized, but this The method has relatively high requirements on tracers and detectors, and it is difficult to popularize

Images